How to use this slides

• Press Esc to navigation mode
• Press pageDown, Right or Down to go to next slide
• Presss PageUP, Left or Up to go to previous slide

Outline

• Recap for observation skills (10 min)
• R Basic (quick recap) and RStudio IDE (10 min)
• Connection (20 min)
  • Crawler’s toolkits in R
• Parsing Data (50 min)
  • XPath & CSS
  • json
  • xml
  • html table
  • regex
• Save your data (10 min)
• Case study
• References

Install: Chrome Extension

• EditThisCookie
• XPath Helper
• JSON Viewer
• Advanced REST client

Recap Observation Skills and HTTP request for Connection

• URL, Method, GET, POST, Headers, Body
• Excample: PTT Gossiping

R Packages Required

Pipeline Coding

• magrittr

Crawler’s toolkits in R

• rvest: a web scraper based on httr and xml2
• httr: toolkit of HTTP methods in R
• XML & : XML parser
• xml2: xml parser based on libxml2

misc

• stringr: string manipulaiton
• data.table: extension of data.frame, a powerful ETL tool in R

Install Packages

(if you haven’t installed those packages yet)

## === install required packages ===
pkg_list <- c("magrittr", "httr", "rvest", "stringr", "data.table",
              "jsonlite", "RSQLite", "devtools")
pkg_new <- pkg_list[!(pkg_list %in% installed.packages()[,"Package"])]
if(length(pkg_new)) install.packages(pkg_new)
if("xmlview" %in% pkg_new) devtools::install_github("hrbrmstr/xmlview")
rm(pkg_new, pkg_list)

Let’s Rock with R!

Hello RStudio

RStudio Settings

Must-known keyboard shortcuts

All RStudio keyboard shortcuts

R recap

• How to get help
• Working Environment
• Basic Data Structure
• Function
• if/else, for, tryCatch
• magrittr
• data.table

How to get help

• ?: Access ducument in R console
• ??: Search the halp system
• google with appropriate keyword. For example:
  • R {package name}
  • R {algorithm name}
• Forum
  • Stack Overflow
    A question and answer site for programmers
  • Taiwan R user Group
    A free R/data analysis user group in Taiwan
  • PTT R_Language board
    A bbs forum for R in Taiwan

Working Environment

Check your working directory everytime you start to work!

Using getwd/setwd to get/set your working directory.

(wd = getwd()) # current config

setwd("resources/img")
getwd()

setwd(wd)

RStudio will set working directory automatically when opening new files.

If you use Projects, RStudio will change working directory for you automatically.

Basic Data Structure

Vector, Matrix, Array, List and Data frame are the most basic data structure in R. These data structures can be mapped into a table according to:

• dimension
• Homogeneous or Heterogeneous of data type

(Atomic) Vector

• (Atomic) vector is a basic data structure in R.
• They are linear vectors of a single primitive type. A scalar is a vector of length 1.

v1 <- c(1:10)
v1
#>  [1]  1  2  3  4  5  6  7  8  9 10
is.vector(v1)
#> [1] TRUE
length(v1)
#> [1] 10
s1 <- 2
s1
#> [1] 2
is.vector(s1)
#> [1] TRUE
length(s1)
#> [1] 1

List

Lists are also vectors, but not atomic vectors. Lists are generic vectors, with (naturally) different semantics.

Elements in a list can be any kinds type and its length is arbitrary.

Function str can help you investigate the structure of a nested list.

li <- list(a = 1:10, 
           b = c("apple", "banana"))
str(li)

#> List of 2
#>  $ a: int [1:10] 1 2 3 4 5 6 7 8 9 10
#>  $ b: chr [1:2] "apple" "banana"

li2 <- list(li = li, 
            c = matrix(1:4, nrow = 2))
str(li2)

#> List of 2
#>  $ li:List of 2
#>   ..$ a: int [1:10] 1 2 3 4 5 6 7 8 9 10
#>   ..$ b: chr [1:2] "apple" "banana"
#>  $ c : int [1:2, 1:2] 1 2 3 4

Visualising lists

x1 <- list(c(1, 2), c(3, 4))
x2 <- list(list(1, 2), list(3, 4))
x3 <- list(1, list(2, list(3)))

Lists Subsetting

a <- list(a = 1:3, b = "a string", c = pi, d = list(-1, -5))

• [ extracts a sub-list. The result will always be a list.

str(a[1:2])
str(a[4])

• [[ extracts a single component from a list. It removes a level of hierarchy from the list.

y <- list("a", 1L, 1.5, TRUE)
str(y[[1]])
str(y[[4]])

• $ is a shorthand for extracting named elements of a list. It works similarly to [[ except that you don’t need to use quotes.

a$a
a[["b"]]

Lists Subsetting (Pepper Shaker)

Data Frame

Data frame is a 2-dimension data structure to deal with a table-like heterogeneous data.

df <- data.frame(gender = c("male", "female", "female", "male"),
                 age = c(33, 18, 24, 26))
## Add new column in a data frame
df$city <- c("Taipei", "Taipei", "Hsinchu", "Taichung")

df

#>   gender age     city
#> 1   male  33   Taipei
#> 2 female  18   Taipei
#> 3 female  24  Hsinchu
#> 4   male  26 Taichung

str(df)

#> 'data.frame':    4 obs. of  3 variables:
#>  $ gender: Factor w/ 2 levels "female","male": 2 1 1 2
#>  $ age   : num  33 18 24 26
#>  $ city  : chr  "Taipei" "Taipei" "Hsinchu" "Taichung"

Recap: data structure

All data structures above are objects. They apply different methods and saved as different type internally.

Function

To understand computations in R, two slogans are helpful:

• Everything that exists is an object.
  
• Everything that happens is a function call.

John Chambers

`+`
#> function (e1, e2)  .Primitive("+")
`<-`
#> .Primitive("<-")
`[`
#> .Primitive("[")
`c`
#> function (..., recursive = FALSE)  .Primitive("c")

Function in R

A typical function in R may look like:

f <- function(par1, par2, ...) {
    # Some magic happened
    return(sth)    # return something
}

• If you don’t use return to specify the return value, the return value will be the last expression inside the function.
• Call by value by default
  send a copy of a object into a function and the input object won’t change after execution without assignment

If

The basc structure of conditional execution in R is:

if (an expression returns TRUE or FALSE) {
    # do something
} else if (another expression returns TRUE or FALSE) {
    # do something
} else {
    # do something
}

for

Iterate items in R.

# iterate a character vector
for (i in c("a", "b")) {
    print(i)
}

#> [1] "a"
#> [1] "b"

# nested loop
m <- matrix(numeric(), nrow = 2, ncol = 2)
for (i in 1:nrow(m)) {
    for (j in 1:ncol(m)) {
        m[i, j] <- i * j
    }
}
m

#>      [,1] [,2]
#> [1,]    1    2
#> [2,]    2    4

tryCatch

• tryCatch let you deal with error handling in R.
• Error handling is IMPORTANT when wrapping your crawler into a function because you don’t want your function fail without any outcome.

tryCatch({
  result <- expr
  # If you want to use more than one 
  # R expression in the "try" part then you'll have to 
  # use curly brackets.
  # 'tryCatch()' will return the last evaluated expression 
  # in case the "try" part was completed successfully
}, warning = function(w) {
  message("Here's the original warning message:")
  message(w)
  # Choose a return value in case of warning
  return(NULL)
}, error = function(e) {
  message("Here's the original error message:")
  message(e)
  # Choose a return value in case of error
  return(NA)
}, finally {
  message("Some other message at the end")
  # finally:
  # Here goes everything that should be executed at the end,
  # regardless of success or error.
  # Usually it's used for releasing resources.
  # If you want more than one expression to be executed, then you 
  # need to wrap them in curly brackets ({...}); otherwise you could
  # just have written 'finally=<expression>' 
})

https://stackoverflow.com/questions/12193779/how-to-write-trycatch-in-r/12195574#12195574

magrittr

What is magrittr

Pipe argument to right-hand side with %>%

• x %>% f is equivalent to f(x)
• x %>% f(y) is equivalent to f(x, y)
• x %>% f %>% g %>% h is equivalent to h(g(f(x)))
• x %>% f(y, .) is equivalent to f(y, x)
• x %>% f(y, z = .) is equivalent to f(y, z = x)

Example

library(magrittr)
iris %>% head()

• Using the dot place-holder

"Ceci n'est pas une pipe" %>% gsub("une", "un", .)

Workflow of Crawler Design

Architecture of Crawlers

Workflow

1. 找到資料頁，想像資料要長什麼樣子，設想產出的資料格式(schema)
2. 觀察網頁內容，找到資料所在的request/response，再一層層往上解析，套上判斷式及迴圈, 完成爬蟲的自動化。
3. 解析取得的資料。

Crawler’s toolkits in R

• rvest: a web scraper based on httr and xml2
• httr: toolkit of HTTP methods in R
• XML & : XML parser
• xml2: xml parser based on libxml2
• RCurl: a wrapper of libcurl

HTTP request

A valid HTTP request includes four things:

• URL: a unique address for a thing
• Method:
  • GET: to retrieve a resource
  • POST: to create a new resource
• Headers: Meta-information about a request
  • User-Agent
  • Cookie
  • Content-type
  • List of HTTP header fields
• Body: data to be send to a server

Web Connector in R

• {httr}
• {RCurl}
• download.file()
• Socket Connection Tools:
  • socketConnection
  • make.socket / read.socket / close.socket

httr

• Basic Features:
  • HTTP verbs: GET(), POST(), HEAD()…
  • http_status: Translate http status code HTTP status code
  • status_code
  • set_cookies(): set cookles
  • add_headers(): add additional headers to a request.
  • headers(): Access response headers
  • content(): Retrieve the contents of a request
    • as = “parsed”: detect content type automatically and parse the result
    • as = “text”: return character
    • as = “raw”: return binary

Connection: GET Method

Use GET() to request data from a specific resource

起手式

## Not Run
library(httr)
res <- GET(
  url = "http://httpbin.org/get",
  add_headers(a = 1, b = 2),
  set_cookies(c = 1, d = 2),
  query = list(q="hihi")
)
content(res, as = "text", encoding = "UTF-8")
content(res, as = "parsed", encoding = "UTF-8")

一個例子學會第一隻爬蟲

• PTT Gossiping

library(magrittr)
library(httr)
library(rvest)
#> Loading required package: xml2

## Connection
url <- "https://www.ptt.cc/bbs/Gossiping/index.html"
res <- GET(url, 
           set_cookies(over18="1"))  # over18 cookie

## Parsing
doc <- res %>% 
  content(as = "text", encoding = "UTF-8") %>% 
  read_html

doc %>% 
  html_nodes(css = ".title a") %>% 
  html_text()
#>  [1] "[問卦] 頭文字D會拍第二集嗎？"                 
#>  [2] "[問卦] 看新唐人電視的都是哪些人?"             
#>  [3] "[新聞] 學當在野黨！藍委「同心營」 選6星渡假村"
#>  [4] "[ＦＢ] 林昶佐接受瑞士SWR電台專訪"             
#>  [5] "[問卦] 寵物帶來樂趣與辛苦度的圖"              
#>  [6] "[問卦] 母性動物護幼子的行為真的很偉大"        
#>  [7] "[問卦] 有沒有一代歌姬白冰冰的八卦"            
#>  [8] "[問卦] 有沒有10次車禍999999999次快的八卦"     
#>  [9] "[問卦] 若沒有8+9的活動可以參加，台灣社會變成?"
#> [10] "[問卦] 全家能買什麼？"                        
#> [11] "[問卦] 有沒有同事愛裝怪聲的八卦？"            
#> [12] "[問卦] 日本人很不耐熱嗎"                      
#> [13] "[新聞] 孔傑榮鼓勵馬英九 訪問中國大陸及美國"   
#> [14] "[問卦] 校園下一個該解禁的項目"                
#> [15] "[公告] 八卦板板規(2016.02.16)"                
#> [16] "[協尋]行車記錄器-5/14承德路5段與基河路交叉口" 
#> [17] "Fw: [申請] 發文門檻提高 5/19-5/26"            
#> [18] "[公告] 五月份置底閒聊文"                      
#> [19] "[公告] 數字類跟風問卦視為鬧板"

Cookies

• A small piece of data sent from a website and stored in the user’s web browser while the user is browsing it
• Someetimes you may need set cookies with set_cookies to retreve a request.

library(httr)
url <- "https://www.ptt.cc/bbs/Gossiping/index.html"
res <- GET(url, 
           set_cookies('over18'='1'))  # over18 cookie

The response status code

url <- "https://www.ptt.cc/bbs/Gossiping/index.html"
res <- GET(url, 
           set_cookies('over18'='1'))  # over18 cookie

# Get an informative description:
http_status(res)
#> $category
#> [1] "Success"
#> 
#> $reason
#> [1] "OK"
#> 
#> $message
#> [1] "Success: (200) OK"

# Or just access the raw code:
res$status_code
#> [1] 200
status_code(res)
#> [1] 200

# highly recommend using one of these functions whenever you're using httr inside a function to make sure you find out about errors as soon as possible.
warn_for_status(res)
stop_for_status(res)

Connection: POST method

• Submits data to be processed to a specified resource
• The query string (name/value pairs) is sent in the HTTP message body

起手式

## Not Run
library(httr)
library(rvest)

res <- POST(url = "http://httpbin.org/post",
             add_headers(a = 1, b = 2),
             set_cookies(c = 1, d = 2),
             body = list(x = "hello", 
                         y = "hihi"))

content(res, as = "text", encoding = "UTF-8")
content(res, as = "parsed", encoding = "UTF-8")

一個例子學會第二隻爬蟲: Guestbook

Guestbook

Exercise: Guestbook

Try to post a message in App Engine GuestBook

Set header

Sometime you may need to provide appropriate HTTP header fields with add_headers() to make a request.

• Starbucks

The Response Body

3 ways to access the body of the request:

The Response Body

httr::content()

res <- GET("https://www.ptt.cc/bbs/NBA/M.1463302851.A.8D7.html")

• binary

(bin <- content(res, as = "raw")) %>% head(200) # raw (binary) vector

#>   [1] 3c 21 44 4f 43 54 59 50 45 20 68 74 6d 6c 3e 0a 3c 68 74 6d 6c 3e 0a
#>  [24] 09 3c 68 65 61 64 3e 0a 09 09 3c 6d 65 74 61 20 63 68 61 72 73 65 74
#>  [47] 3d 22 75 74 66 2d 38 22 3e 0a 09 09 0a 0a 3c 6d 65 74 61 20 6e 61 6d
#>  [70] 65 3d 22 76 69 65 77 70 6f 72 74 22 20 63 6f 6e 74 65 6e 74 3d 22 77
#>  [93] 69 64 74 68 3d 64 65 76 69 63 65 2d 77 69 64 74 68 2c 20 69 6e 69 74
#> [116] 69 61 6c 2d 73 63 61 6c 65 3d 31 22 3e 0a 0a 3c 74 69 74 6c 65 3e 5b
#> [139] e8 a8 8e e8 ab 96 5d 20 e6 9e 97 e6 9b b8 e8 b1 aa e5 a6 82 e6 9e 9c
#> [162] e5 8a a0 e5 85 a5 e9 a6 ac e5 88 ba e8 83 bd e5 be 97 e5 88 b0 e5 a4
#> [185] 9a e5 b0 91 e4 b8 8a e5 a0 b4 e6 99 82 e9 96 93

# writeBin(bin, "myfile.txt") # the highest fidelity way of saving files to disk

• text

content(res, as = "text", encoding = "UTF-8") %>%  # accesses the body as a character vector
  `Encoding<-`("UTF-8")

#> [1] "<!DOCTYPE html>\n<html>\n\t<head>\n\t\t<meta charset=\"utf-8\">\n\t\t\n\n<meta name=\"viewport\" content=\"width=device-width, initial-scale=1\">\n\n<title>[討論] 林書豪如果加入馬刺能得到多少上場時間? - 看板 NBA - 批踢踢實業坊</title>\n<meta name=\"robots\" content=\"all\">\n<meta name=\"keywords\" content=\"Ptt BBS 批踢踢\">\n<meta name=\"description\" content=\"\n\n\n最近有傳言說林書豪可能會加入馬刺\n\n\">\n<meta property=\"og:site_name\" content=\"Ptt 批踢踢實業坊\">\n<meta property=\"og:title\" content=\"[討論] 林書豪如果加入馬刺能得到多少上場時間?\">\n<meta property=\"og:description\" content=\"\n\n\n最近有傳言說林書豪可能會加入馬刺\n\n\">\n<link rel=\"canonical\" href=\"https://www.ptt.cc/bbs/NBA/M.1463302851.A.8D7.html\">\n\n<link rel=\"stylesheet\" type=\"text/css\" href=\"//images.ptt.cc/v2.17/bbs-common.css\">\n<link rel=\"stylesheet\" type=\"text/css\" href=\"//images.ptt.cc/v2.17/bbs-base.css\" media=\"screen\">\n<link rel=\"stylesheet\" type=\"text/css\" href=\"//images.ptt.cc/v2.17/bbs-custom.css\">\n<link rel=\"stylesheet\" type=\"text/css\" href=\"//images.ptt.cc/v2.17/pushstream.css\" media=\"screen\">\n<link rel=\"stylesheet\" type=\"text/css\" href=\"//images.ptt.cc/v2.17/bbs-print.css\" media=\"print\">\n\n\n<script src=\"//ajax.googleapis.com/ajax/libs/jquery/2.1.1/jquery.min.js\"></script>\n<script src=\"//images.ptt.cc/v2.17/bbs.js\"></script>\n\n\n\t\t\n\n<script type=\"text/javascript\">\n\n  var _gaq = _gaq || [];\n  _gaq.push(['_setAccount', 'UA-32365737-1']);\n  _gaq.push(['_setDomainName', 'ptt.cc']);\n  _gaq.push(['_trackPageview']);\n\n  (function() {\n    var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true;\n    ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js';\n    var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s);\n  })();\n\n</script>\n\n\n\t</head>\n    <body>\n\t\t\n<div id=\"fb-root\"></div>\n<script>(function(d, s, id) {\nvar js, fjs = d.getElementsByTagName(s)[0];\nif (d.getElementById(id)) return;\njs = d.createElement(s); js.id = id;\njs.src = \"//connect.facebook.net/en_US/all.js#xfbml=1\";\nfjs.parentNode.insertBefore(js, fjs);\n}(document, 'script', 'facebook-jssdk'));</script>\n\n<div id=\"topbar-container\">\n\t<div id=\"topbar\" class=\"bbs-content\">\n\t\t<a id=\"logo\" href=\"/\">批踢踢實業坊</a>\n\t\t<span>&rsaquo;</span>\n\t\t<a class=\"board\" href=\"/bbs/NBA/index.html\"><span class=\"board-label\">看板 </span>NBA</a>\n\t\t<a class=\"right small\" href=\"/about.html\">關於我們</a>\n\t\t<a class=\"right small\" href=\"/contact.html\">聯絡資訊</a>\n\t</div>\n</div>\n<div id=\"navigation-container\">\n\t<div id=\"navigation\" class=\"bbs-content\">\n\t\t<a class=\"board\" href=\"/bbs/NBA/index.html\">返回看板</a>\n\t\t<div class=\"bar\"></div>\n\t\t<div class=\"share\">\n\t\t\t<span>分享</span>\n\t\t\t<div class=\"fb-like\" data-send=\"false\" data-layout=\"button_count\" data-width=\"90\" data-show-faces=\"false\" data-href=\"http://www.ptt.cc/bbs/NBA/M.1463302851.A.8D7.html\"></div>\n\n\t\t\t<div class=\"g-plusone\" data-size=\"medium\"></div>\n<script type=\"text/javascript\">\nwindow.___gcfg = {lang: 'zh-TW'};\n(function() {\nvar po = document.createElement('script'); po.type = 'text/javascript'; po.async = true;\npo.src = 'https://apis.google.com/js/plusone.js';\nvar s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(po, s);\n})();\n</script>\n\n\t\t</div>\n\t</div>\n</div>\n<div id=\"main-container\">\n    <div id=\"main-content\" class=\"bbs-screen bbs-content\"><div class=\"article-metaline\"><span class=\"article-meta-tag\">作者</span><span class=\"article-meta-value\">magneto5566 (萬磁王5566_麥寮法斯賓達)</span></div><div class=\"article-metaline-right\"><span class=\"article-meta-tag\">看板</span><span class=\"article-meta-value\">NBA</span></div><div class=\"article-metaline\"><span class=\"article-meta-tag\">標題</span><span class=\"article-meta-value\">[討論] 林書豪如果加入馬刺能得到多少上場時間?</span></div><div class=\"article-metaline\"><span class=\"article-meta-tag\">時間</span><span class=\"article-meta-value\">Sun May 15 17:00:48 2016</span></div>\n\n\n最近有傳言說林書豪可能會加入馬刺\n\n不管對林書豪或對球迷來說馬刺都是個很棒的選擇\n\n馬刺今年例行賽戰績高達67勝,主力陣容維持的話明年應該依舊有60勝左右\n\n雖然波波維奇用兵如神,但林書豪在馬刺到底能獲得多少上場時間呢\n\n林書豪應該是替補PG和SG的位置\n\n單看馬刺和雷霆的系列賽\n\n馬刺的後場部分主要由4位球員分擔時間\n\n先發: Danny Green , Tony Parker\n\n替補: Manu Ginobili, Patty Mills\n\n\n理論上 林書豪最有機會的是Ginobili的位置\n\n畢竟鬼切已經38歲,下季會不會回來都還不確定\n\n林書豪的切入能力雖然沒有巔峰的鬼切犀利\n\n但是大勝38歲鬼切應該是沒問題\n\n相信波波維奇很了解這一點\n\n不知道大家怎麼看呢?\n\n\n--\n<span class=\"f2\">※ 發信站: 批踢踢實業坊(ptt.cc), 來自: 140.112.71.9\n</span><span class=\"f2\">※ 文章網址: <a href=\"https://www.ptt.cc/bbs/NBA/M.1463302851.A.8D7.html\" target=\"_blank\" rel=\"nofollow\">https://www.ptt.cc/bbs/NBA/M.1463302851.A.8D7.html</a>\n</span><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">sss1234     </span><span class=\"f3 push-content\">: 87分鐘</span><span class=\"push-ipdatetime\"> 05/15 17:02\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">h22212247888</span><span class=\"f3 push-content\">: 8.7分鐘</span><span class=\"push-ipdatetime\"> 05/15 17:02\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">rick        </span><span class=\"f3 push-content\">: 等七月再來分析好嗎?</span><span class=\"push-ipdatetime\"> 05/15 17:02\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">demonh311   </span><span class=\"f3 push-content\">: 當然是當基石</span><span class=\"push-ipdatetime\"> 05/15 17:02\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">arcss       </span><span class=\"f3 push-content\">: 前半年冰凍，後半年15分鐘/場</span><span class=\"push-ipdatetime\"> 05/15 17:03\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">hunt5566    </span><span class=\"f3 push-content\">: 呵呵 波波又藏招不讓豪鬼上囉</span><span class=\"push-ipdatetime\"> 05/15 17:04\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">Xenogamer   </span><span class=\"f3 push-content\">: 0</span><span class=\"push-ipdatetime\"> 05/15 17:04\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">love1500274 </span><span class=\"f3 push-content\">: Pop會讓書豪取代跑車先發 書豪會拿下年度mvp</span><span class=\"push-ipdatetime\"> 05/15 17:05\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">littlegreen </span><span class=\"f3 push-content\">: 馬刺需要切入沒錯 但是切傳能力 這書豪比較弱</span><span class=\"push-ipdatetime\"> 05/15 17:06\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">噓 </span><span class=\"f3 hl push-userid\">r30385      </span><span class=\"f3 push-content\">: 8.7秒</span><span class=\"push-ipdatetime\"> 05/15 17:08\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">donnylee    </span><span class=\"f3 push-content\">: Mills也差不多要退了，40歲</span><span class=\"push-ipdatetime\"> 05/15 17:12\n</span></div>         Mills 27歲而已\n<span class=\"f2\">※ 編輯: magneto5566 (140.112.71.9), 05/15/2016 17:14:25\n</span><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">darkfeather </span><span class=\"f3 push-content\">: 美媒建議變傳言，這進化速度還真快</span><span class=\"push-ipdatetime\"> 05/15 17:15\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">donnylee    </span><span class=\"f3 push-content\">: 喔，我記錯了，那是哪一個40？</span><span class=\"push-ipdatetime\"> 05/15 17:18\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">Mooooose    </span><span class=\"f3 push-content\">: Popo黑掉的所花的時間會比較多嗎?</span><span class=\"push-ipdatetime\"> 05/15 17:18\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">erotica     </span><span class=\"f3 push-content\">: 阿豪打1 2號都可以  還可以各種大小鎖防守</span><span class=\"push-ipdatetime\"> 05/15 17:21\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">噓 </span><span class=\"f3 hl push-userid\">king181239  </span><span class=\"f3 push-content\">: 去專版問啊</span><span class=\"push-ipdatetime\"> 05/15 17:21\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">king181239  </span><span class=\"f3 push-content\">: 他們會說超過35分鐘</span><span class=\"push-ipdatetime\"> 05/15 17:21\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">erotica     </span><span class=\"f3 push-content\">: 屌打上述4個人</span><span class=\"push-ipdatetime\"> 05/15 17:21\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">king181239  </span><span class=\"f3 push-content\">: 事實上只能出賽20幾</span><span class=\"push-ipdatetime\"> 05/15 17:22\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">噓 </span><span class=\"f3 hl push-userid\">s925407     </span><span class=\"f3 push-content\">: 怎麼會有人說Mills40歲啦，一日球迷也沒這麼不專業</span><span class=\"push-ipdatetime\"> 05/15 17:22\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">s925407     </span><span class=\"f3 push-content\">: 啦</span><span class=\"push-ipdatetime\"> 05/15 17:22\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">a10141013   </span><span class=\"f3 push-content\">: 馬刺大約20-25分鐘 然後popo被說冰箱</span><span class=\"push-ipdatetime\"> 05/15 17:24\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">dream1285201</span><span class=\"f3 push-content\">: 一個球員會黑掉大多跟他的球迷有關</span><span class=\"push-ipdatetime\"> 05/15 17:24\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">smtp        </span><span class=\"f3 push-content\">: 重點是馬刺想出多少爭取豪哥?</span><span class=\"push-ipdatetime\"> 05/15 17:28\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">kl015013    </span><span class=\"f3 push-content\">: 20</span><span class=\"push-ipdatetime\"> 05/15 17:29\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">bycarbird   </span><span class=\"f3 push-content\">: 0，因為書豪不是波波的菜</span><span class=\"push-ipdatetime\"> 05/15 17:30\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">MoMovincent </span><span class=\"f3 push-content\">: 那個說40歲的應該是要說a米吧</span><span class=\"push-ipdatetime\"> 05/15 17:30\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">s95115260   </span><span class=\"f3 push-content\">: 馬刺怎麼可能要球權的球員</span><span class=\"push-ipdatetime\"> 05/15 17:31\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">love1500274 </span><span class=\"f3 push-content\">: 他想說的是Miller 打成Mills  lol</span><span class=\"push-ipdatetime\"> 05/15 17:32\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">kiwi1220    </span><span class=\"f3 push-content\">: 40是 ，Miller</span><span class=\"push-ipdatetime\"> 05/15 17:33\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">miler22020  </span><span class=\"f3 push-content\">: 20分鐘 球權不多 然後popo從此黑掉</span><span class=\"push-ipdatetime\"> 05/15 17:36\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">scatology   </span><span class=\"f3 push-content\">: 5 垃圾時間</span><span class=\"push-ipdatetime\"> 05/15 17:36\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">smtp        </span><span class=\"f3 push-content\">: 10M簽的話, 上場時間不可能只有20分鐘...</span><span class=\"push-ipdatetime\"> 05/15 17:37\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">forker561   </span><span class=\"f3 push-content\">: 第一年不會有上場時間 最快第二年才有上場時間</span><span class=\"push-ipdatetime\"> 05/15 17:38\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">candy9999   </span><span class=\"f3 push-content\">: 去哪都比在黃蜂好</span><span class=\"push-ipdatetime\"> 05/15 17:40\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">DiAbLoE     </span><span class=\"f3 push-content\">: popo會變黑人</span><span class=\"push-ipdatetime\"> 05/15 17:41\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">smtp        </span><span class=\"f3 push-content\">: 豪哥今年曾說,年薪影響別人對你評價,可見企圖心很強</span><span class=\"push-ipdatetime\"> 05/15 17:41\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">happyalex   </span><span class=\"f3 push-content\">: 公道價8分鐘</span><span class=\"push-ipdatetime\"> 05/15 17:43\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">goury       </span><span class=\"f3 push-content\">: 千萬不要去馬刺...Linsanity的重點是明星度，而不是</span><span class=\"push-ipdatetime\"> 05/15 17:47\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">goury       </span><span class=\"f3 push-content\">: 馬刺這樣只在乎勝利，讓買票進場的人看不到球星的隊</span><span class=\"push-ipdatetime\"> 05/15 17:48\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">loverxa     </span><span class=\"f3 push-content\">: 到時不用豪神 就沒人在跟你用兵如神了啦</span><span class=\"push-ipdatetime\"> 05/15 17:49\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">crazywill   </span><span class=\"f3 push-content\">: 馬刺不會要阿豪 不用討論了</span><span class=\"push-ipdatetime\"> 05/15 17:50\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">birdman5656 </span><span class=\"f3 push-content\">: 先等波波帶完奧運吧</span><span class=\"push-ipdatetime\"> 05/15 17:51\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">yyyyy       </span><span class=\"f3 push-content\">: 馬刺應該以我豪為核心進行重建才對</span><span class=\"push-ipdatetime\"> 05/15 17:53\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">leocheng    </span><span class=\"f3 push-content\">: 馬刺喜歡跳投好的 怎麼可能愛豪鬼</span><span class=\"push-ipdatetime\"> 05/15 18:04\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">june0204    </span><span class=\"f3 push-content\">: 還是2000找康利吧</span><span class=\"push-ipdatetime\"> 05/15 18:12\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">HowWhy99    </span><span class=\"f3 push-content\">: 8.7分 不能再多了</span><span class=\"push-ipdatetime\"> 05/15 18:12\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">crazylin924 </span><span class=\"f3 push-content\">: 35分鐘 （一整季</span><span class=\"push-ipdatetime\"> 05/15 18:19\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">噓 </span><span class=\"f3 hl push-userid\">xman262     </span><span class=\"f3 push-content\">: 487秒</span><span class=\"push-ipdatetime\"> 05/15 18:24\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">噓 </span><span class=\"f3 hl push-userid\">ppo7741     </span><span class=\"f3 push-content\">: 8.7分，跟這篇文的分數一樣，不能再高了</span><span class=\"push-ipdatetime\"> 05/15 18:25\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">carey1119   </span><span class=\"f3 push-content\">: 8.7分鐘</span><span class=\"push-ipdatetime\"> 05/15 18:33\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">DurantKevin </span><span class=\"f3 push-content\">: 。。。</span><span class=\"push-ipdatetime\"> 05/15 18:34\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">qq1029qq    </span><span class=\"f3 push-content\">: 87分鐘無誤</span><span class=\"push-ipdatetime\"> 05/15 18:40\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">drew9992    </span><span class=\"f3 push-content\">: <a href=\"http://i.imgur.com/0XBGyQU.jpg\" target=\"_blank\" rel=\"nofollow\">http://i.imgur.com/0XBGyQU.jpg</a></span><span class=\"push-ipdatetime\"> 05/15 19:01\n</span></div><div class=\"richcontent\"><img src=\"//i.imgur.com/0XBGyQU.jpg\" alt=\"\" /></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">dai26       </span><span class=\"f3 push-content\">: 若能完全取代馬妞,大概30分鐘，而且可打滿第四節</span><span class=\"push-ipdatetime\"> 05/15 19:20\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">vicyong     </span><span class=\"f3 push-content\">: 不要在幻想了</span><span class=\"push-ipdatetime\"> 05/15 19:23\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">gotohikaru  </span><span class=\"f3 push-content\">: 第一年不用想</span><span class=\"push-ipdatetime\"> 05/15 19:32\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">waw20002002 </span><span class=\"f3 push-content\">: 8.7分鐘！不能在多了</span><span class=\"push-ipdatetime\"> 05/15 20:13\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">噓 </span><span class=\"f3 hl push-userid\">an565an565  </span><span class=\"f3 push-content\">: 後衛來馬刺都要被冰一整季</span><span class=\"push-ipdatetime\"> 05/15 21:07\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">trylovetom  </span><span class=\"f3 push-content\">: Parker不是被豪哥哥電？</span><span class=\"push-ipdatetime\"> 05/15 21:09\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">smtp        </span><span class=\"f3 push-content\">: Parker是輸在年紀, 巔峰時不可能會被豪哥電~</span><span class=\"push-ipdatetime\"> 05/15 21:12\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">噓 </span><span class=\"f3 hl push-userid\">Yginger1    </span><span class=\"f3 push-content\">: 42分鐘 全場打好打滿</span><span class=\"push-ipdatetime\"> 05/15 21:19\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">pickoff     </span><span class=\"f3 push-content\">: 馬刺不收。問下一隊謝謝</span><span class=\"push-ipdatetime\"> 05/15 21:22\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">mmshlovebob </span><span class=\"f3 push-content\">: 8.7分鐘 不能再多了</span><span class=\"push-ipdatetime\"> 05/15 21:36\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">roc074      </span><span class=\"f3 push-content\">: 8.7時</span><span class=\"push-ipdatetime\"> 05/15 22:04\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">噓 </span><span class=\"f3 hl push-userid\">deathgale   </span><span class=\"f3 push-content\">: 等加盟了再來討論，記者都自以為球團老闆...</span><span class=\"push-ipdatetime\"> 05/15 22:38\n</span></div><div class=\"push\"><span class=\"hl push-tag\">推 </span><span class=\"f3 hl push-userid\">squall410339</span><span class=\"f3 push-content\">: 快去啦馬刺黑掉就是爽</span><span class=\"push-ipdatetime\"> 05/15 23:29\n</span></div><div class=\"push\"><span class=\"f1 hl push-tag\">→ </span><span class=\"f3 hl push-userid\">raku        </span><span class=\"f3 push-content\">: 一開始一場應該是不到10分鐘...</span><span class=\"push-ipdatetime\"> 05/16 00:34\n</span></div></div>\n    \n    <div id=\"article-polling\" data-pollurl=\"/poll/NBA/M.1463302851.A.8D7.html?cacheKey=2052-14787853&offset=7844&offset-sig=05b0d881a2af63a68de9ce2d0f1cda56eec2d548\" data-longpollurl=\"/v1/longpoll?id=5ef93f8aa82d4cf930eb27310d147571cf13936a\" data-offset=\"7844\"></div>\n    \n\n    \n</div>\n\n    </body>\n</html>\n"

• auto parsed
  httr provides a number of default parsers for common file types:
  html_document, csv, json, jpeg, png. See ?httr::content

content(res, as = "parsed")

#> {xml_document}
#> <html>
#> [1] <head>\n\t\t<meta charset="utf-8"/>\n\t\t\n\n<meta name="viewport" c ...
#> [2] <body>\n\t\t\n<div id="fb-root"/>\n<script><![CDATA[(function(d, s,  ...

Beware of System Encoding

• Parse data with appropriate encoding
• Use Encoding(), iconv() and Sys.setlocale() in Windows (non-UTF8 system).
• iconvlist() to show all encodings availabe.

Encoding

?Encoding ?iconv

• Unix-like

## check out your system locale
Sys.getlocale()
#> [1] "en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8"
aa <- "你好嗎"
Encoding(aa)
#> [1] "UTF-8"
charToRaw(aa)
#> [1] e4 bd a0 e5 a5 bd e5 97 8e

(aa_big5 <- iconv(aa, from = "UTF-8", to = "Big5"))
#> [1] "\xa7A\xa6n\xb6\xdc"
Encoding(aa_big5)
#> [1] "unknown"
harToRaw(aa_big5)
#> [1] a7 41 a6 6e b6 dc

• Windows

# On Windows
Sys.getlocale()
#> "CP950"
aa <- "你好嗎"
Encoding(aa)
#> "unknown"
aa_utf8 <- iconv(aa, from = "Big5", to = "UTF-8")
Encoding(aa_utf8)
#> "UTF-8"

The secret of URL

URL?par1=val1&par2=val2

• The query string is sent in the URL of a GET request
• Non-ASCII character and some preserved characters need to be encoded with URL Encode
• In most situation, you may need to manipulate the query string when you write a crawler

Query String

You can assign query parameter with query()

• Example: http://ecshweb.pchome.com.tw/search/v3.3/?q=sony

res1 <- GET(
  "http://ecshweb.pchome.com.tw/search/v3.3/all/results?q=sony&page=1&sort=rnk/dc"
)

res2 <- GET("http://ecshweb.pchome.com.tw/search/v3.3/all/results",
            query = list(q="sony", page="1", sort="rnk/dc"))

Url Encoding in R

URLencode(" ")

#> [1] "%20"

greeting = "你好嗎我很好"
greeting_enc = URLencode(greeting)
greeting_enc

#> [1] "%E4%BD%A0%E5%A5%BD%E5%97%8E%E6%88%91%E5%BE%88%E5%A5%BD"

URLdecode(greeting_enc)

#> [1] "你好嗎我很好"

Concatenate strings / String formatting

paste0("hihi", greeting)

#> [1] "hihi你好嗎我很好"

paste0("hihi", greeting, 1:3)

#> [1] "hihi你好嗎我很好1" "hihi你好嗎我很好2" "hihi你好嗎我很好3"

paste("hihi", greeting, 1:3, sep = " ", collapse = ",")

#> [1] "hihi 你好嗎我很好 1,hihi 你好嗎我很好 2,hihi 你好嗎我很好 3"

sprintf("%s,%s嗎?", "hihi", greeting)

#> [1] "hihi,你好嗎我很好嗎?"

What to Parse

結構化資料

• html elements: (rvest::html_node)
• json, jsonp: PChome, 紫外線即時監測資料 (jsonlite::fromJSON)
• html table: 公開資訊觀測站 (rvest::html_table, XML::readHtmlTable)
• xml table: 7-11 (XML::xmlToDataFrame)

非結構化資料

• RegEx: 7-11 的 citycode (areacode.js)、東森房屋

Parsing Response Content: Text Type

rvest

A web scraper designed to work with magrittr.

• Create a html document with read_html()
• Select parts of a document using css selectors or xpath
  • html_nodes(doc, css = "<css selector>")
  • html_nodes(doc, xpath = "<css selector>")
• Extract components with
  • html_name(): the name of the tag
  • html_text(): all text inside the tag
  • html_attr(): contents of a single attribute
  • html_attrs(): all attributes
  • html_table: parse tables into data frames

Use with rvest

• 起手式

GET(url) %>% content(as="text") %>% rvest::read_html()

Tree Structure of HTML (Document Object Model, DOM)

<a href = "www.meetup.com/Taiwan-R">
  Taiwan R User Group Website
</a>

• nodename : a
• attribute : href with value "www.meetup.com/Taiwan-R"
• text: Taiwan R User Group Website

A simple HTML document

demo_page

library(magrittr)
doc = readLines("http://leoluyi.github.io/RCrawler101_201605_Week2/resources/data/demo.html") %>%
    paste(collapse = "\n")
cat(doc)

#> <!DOCTYPE HTML>
#> <html lang="zh-TW">
#> <head>
#>     <meta charset="UTF-8">
#>     <title>Document</title>
#> </head>
#> <html>
#>     <body>
#>         <div id='title' class='character'>
#>             <a class="link" href="http://data-sci.info/r-crawler-101/">data-sci.info</a>
#>         </div>
#>         <div id='summary' class='character'>
#>             <span class='title'>Number:</span>
#>             <span class='number'>2</span>
#>         </div>
#>         <div id='table1' class='info'>
#>             <table style="border-collapse: collapse; border: 1px solid black;">
#>                 <tr>
#>                     <th>Name</th>
#>                     <th>Gender</th>
#>                     <th>Age</th>
#>                 </tr>
#>                 <tr>
#>                     <td>Alice</td>
#>                     <td>Female</td>
#>                     <td>24</td>
#>                 </tr>
#>                 <tr>
#>                     <td>Jane</td>
#>                     <td>Female</td>
#>                     <td>26</td>
#>                 </tr>
#>             </table>
#>         </div>
#> 
#>         <h3>傷心排行榜</h3>
#>         <div id='table2' class='info'>
#>             <table style="border-collapse: collapse; border: 1px solid black;">
#>                 <tr>
#>                     <th>姓名</th>
#>                     <th>年齡</th>
#>                 </tr>
#>                 <tr>
#>                     <td>白素貞</td>
#>                     <td>15</td>
#>                 </tr>
#>                 <tr>
#>                     <td>聶小倩</td>
#>                     <td>32</td>
#>                 </tr>
#>                 <tr>
#>                     <td>祝英台</td>
#>                     <td>54</td>
#>                 </tr>
#>             </table>
#>         </div>
#>     </body>
#> </html>

CSS practice
http://flukeout.github.io/

doc <- read_html("http://leoluyi.github.io/RCrawler101_201605_Week2/resources/data/demo.html")
doc %>% 
    html_nodes(css = ".character") %>% 
    html_text

#> [1] "\n            data-sci.info\n        "         
#> [2] "\n            Number:\n            2\n        "

doc %>% 
    html_nodes(css = "#title > .link") %>% 
    html_text

#> [1] "data-sci.info"

Extract with xpath

doc <- read_html("http://leoluyi.github.io/RCrawler101_201605_Week2/resources/data/demo.html")
doc %>% 
    html_nodes(xpath = "//*[@class='character']") %>% 
    html_text

#> [1] "\n            data-sci.info\n        "         
#> [2] "\n            Number:\n            2\n        "

doc %>% 
    html_nodes(xpath = "//div[@id='title']/a") %>% 
    html_text

#> [1] "data-sci.info"

Extract name of tag

node = doc %>% 
    html_nodes(css = "#summary") %>% 
    html_name
node

#> [1] "div"

Extract link

link = doc %>% 
    html_nodes(xpath = "/html/body/div[@id='title']/a") %>%
    html_attr("href")
link

#> [1] "http://data-sci.info/r-crawler-101/"

Example: PTT

use package xmlview to parse XML content and extract elements.

• PTT

Extract table

rvest::html_table()

(html_table() is temperarily not working with non-ascii data on Windows.)

students <- read_html("http://leoluyi.github.io/RCrawler101_201605_Week2/resources/data/demo.html") %>% 
    html_nodes(css = "#table1>table") %>%
    html_table()
students

#> [[1]]
#>    Name Gender Age
#> 1 Alice Female  24
#> 2  Jane Female  26

Extract table

XML::readHTMLTable()

demo_page

res_text <- GET("http://leoluyi.github.io/RCrawler101_201605_Week2/resources/data/demo.html") %>% 
  content("text", encoding = "UTF-8") %>% 
  `Encoding<-`("UTF-8")

# read and extract tables from html string
tables <- res_text %>% XML::readHTMLTable(encoding = "UTF-8")

str(tables) # take a look 
View(tables[[2]]) # take a look 

• Yahoo Stock

Example: 公開資訊觀測站

• 公開資訊觀測站

Parse XML table

Parse XML table with XML::xmlToDataFrame()

Example

• Seven Eleven
• RegEx: 7-11 citycode (areacode.js)

Parse JSON format

Yun can parse JSON with jsonlite in R

Parse JSON with jsonlite

• Example: PChome API

library(jsonlite)
library(magrittr)
res <- GET("http://ecshweb.pchome.com.tw/search/v3.3/all/results?q=sony&page=1&sort=rnk/dc")

• jsonlite::fromJSON()

res_df <- content(res, as = "text", encoding = "UTF-8") %>% 
  fromJSON() %>%
  .$prods  # equivelent to (function(x) {x$prods})

head(res_df)
#>                 Id cateId
#> 1 DYAD2P-A9006JOZX DYAD2P
#> 2 DYAD2R-A9006XKSI DYAD2R
#> 3 DPAE03-A9006RR9X DPAE0V
#> 4 DYAD2O-A9006JDQE DYAD2O
#> 5 DYAD2O-A9006J87V DYAD2O
#> 6 DGBT30-A900649ND DYAQA6
#>                                                                          picS
#> 1 /pic/v1/data/item/201510/D/Y/A/D/2/P/sDYAD2P-A9006JOZX000_5614b4cc793f6.jpg
#> 2 /pic/v1/data/item/201604/D/Y/A/D/2/R/sDYAD2R-A9006XKSI000_56fe3f0b20c2e.jpg
#> 3 /pic/v1/data/item/201601/D/P/A/E/0/3/sDPAE03-A9006RR9X000_56970708b9924.jpg
#> 4 /pic/v1/data/item/201510/D/Y/A/D/2/O/sDYAD2O-A9006JDQE000_560e037b08db2.jpg
#> 5 /pic/v1/data/item/201509/D/Y/A/D/2/O/sDYAD2O-A9006J87V000_560b9b7cb48ad.jpg
#> 6 /pic/v1/data/item/201505/D/G/B/T/3/0/sDGBT30-A900649ND000_5567e78207366.jpg
#>                                                                         picB
#> 1 /pic/v1/data/item/201510/D/Y/A/D/2/P/DYAD2P-A9006JOZX000_5614b4cc76ca5.jpg
#> 2 /pic/v1/data/item/201605/D/Y/A/D/2/R/DYAD2R-A9006XKSI000_57426fcf0b981.jpg
#> 3 /pic/v1/data/item/201605/D/P/A/E/0/3/DPAE03-A9006RR9X000_573e6e365d2c4.jpg
#> 4 /pic/v1/data/item/201604/D/Y/A/D/2/O/DYAD2O-A9006JDQE000_57198b2dc111b.jpg
#> 5 /pic/v1/data/item/201605/D/Y/A/D/2/O/DYAD2O-A9006J87V000_57426e333e83e.jpg
#> 6 /pic/v1/data/item/201604/D/G/B/T/3/0/DGBT30-A900649ND000_5715927991a26.jpg
#>                                           name
#> 1 Sony Xperia Z5 Compact E5823 4.6吋八核輕旗艦
#> 2                       SONY Xperia Z5 Premium
#> 3                   Sony 行動微型投影機 MP-CL1
#> 4                               SONY Xperia Z5
#> 5                               SONY Xperia Z5
#> 6            SONY SBH70 (防水IP57)無線藍牙耳機
#>                                                           describe price
#> 1 送9H玻璃保貼+保護套 Sony Xperia Z5 Compact E5823 4.6吋八核輕旗艦 13880
#> 2                   ▼送32G卡+手機立架+玻璃貼SONY Xperia Z5 Premium 19900
#> 3                                       Sony 行動微型投影機 MP-CL1 11900
#> 4 超值▼送行動電源+保護貼+保護套▼SONY Xperia Z5 5.2吋美型防水旗艦機 16888
#> 5                           ★送保貼+支架+觸控筆吊飾★SONY Xperia Z5 16990
#> 6                                SONY SBH70 (防水IP57)無線藍牙耳機  2188
#>   author brand publishDate isPick
#> 1                               0
#> 2                               0
#> 3                               0
#> 4                               0
#> 5                               0
#> 6                               0

httr auto parse json as list

res_list <- content(res, as = "parsed")
# str(res_list)
res_list$prods[[1]]

#> $Id
#> [1] "DYAD2P-A9006JOZX"
#> 
#> $cateId
#> [1] "DYAD2P"
#> 
#> $picS
#> [1] "/pic/v1/data/item/201510/D/Y/A/D/2/P/sDYAD2P-A9006JOZX000_5614b4cc793f6.jpg"
#> 
#> $picB
#> [1] "/pic/v1/data/item/201510/D/Y/A/D/2/P/DYAD2P-A9006JOZX000_5614b4cc76ca5.jpg"
#> 
#> $name
#> [1] "Sony Xperia Z5 Compact E5823 4.6吋八核輕旗艦"
#> 
#> $describe
#> [1] "送9H玻璃保貼+保護套 Sony Xperia Z5 Compact E5823 4.6吋八核輕旗艦"
#> 
#> $price
#> [1] 13880
#> 
#> $author
#> [1] ""
#> 
#> $brand
#> [1] ""
#> 
#> $publishDate
#> [1] ""
#> 
#> $isPick
#> [1] 0

Combind list of dataframes with loop

# str(res_list$prods)
res_df2 = data.frame()
for (i in 1:length(res_list$prods)) {
  res_df2 = rbind(res_df2, 
                  data.frame(res_list$prods[[i]], 
                             stringsAsFactors = FALSE))
}
identical(res_df, res_df2)

#> [1] TRUE

(better use do.call + rbind)

More general way to combine list of dataframes

• do.call + rbind
• data.table::rbindlist

Parse Parse list of dataframes with

do.call + rbind

# make a list of dataframes
df_list <- lapply(res_list$prods, as.data.frame, stringsAsFactors = FALSE)

# combine dataframes
res_df3 = do.call(rbind, df_list)
identical(res_df, res_df3)

#> [1] TRUE

# str(res_df3)

data.table::rbindlist

library(data.table)
res_df4 = rbindlist(df_list)
head(res_df4)

#>                  Id cateId
#> 1: DYAD2P-A9006JOZX DYAD2P
#> 2: DYAD2R-A9006XKSI DYAD2R
#> 3: DPAE03-A9006RR9X DPAE0V
#> 4: DYAD2O-A9006JDQE DYAD2O
#> 5: DYAD2O-A9006J87V DYAD2O
#> 6: DGBT30-A900649ND DYAQA6
#>                                                                           picS
#> 1: /pic/v1/data/item/201510/D/Y/A/D/2/P/sDYAD2P-A9006JOZX000_5614b4cc793f6.jpg
#> 2: /pic/v1/data/item/201604/D/Y/A/D/2/R/sDYAD2R-A9006XKSI000_56fe3f0b20c2e.jpg
#> 3: /pic/v1/data/item/201601/D/P/A/E/0/3/sDPAE03-A9006RR9X000_56970708b9924.jpg
#> 4: /pic/v1/data/item/201510/D/Y/A/D/2/O/sDYAD2O-A9006JDQE000_560e037b08db2.jpg
#> 5: /pic/v1/data/item/201509/D/Y/A/D/2/O/sDYAD2O-A9006J87V000_560b9b7cb48ad.jpg
#> 6: /pic/v1/data/item/201505/D/G/B/T/3/0/sDGBT30-A900649ND000_5567e78207366.jpg
#>                                                                          picB
#> 1: /pic/v1/data/item/201510/D/Y/A/D/2/P/DYAD2P-A9006JOZX000_5614b4cc76ca5.jpg
#> 2: /pic/v1/data/item/201605/D/Y/A/D/2/R/DYAD2R-A9006XKSI000_57426fcf0b981.jpg
#> 3: /pic/v1/data/item/201605/D/P/A/E/0/3/DPAE03-A9006RR9X000_573e6e365d2c4.jpg
#> 4: /pic/v1/data/item/201604/D/Y/A/D/2/O/DYAD2O-A9006JDQE000_57198b2dc111b.jpg
#> 5: /pic/v1/data/item/201605/D/Y/A/D/2/O/DYAD2O-A9006J87V000_57426e333e83e.jpg
#> 6: /pic/v1/data/item/201604/D/G/B/T/3/0/DGBT30-A900649ND000_5715927991a26.jpg
#>                                            name
#> 1: Sony Xperia Z5 Compact E5823 4.6吋八核輕旗艦
#> 2:                       SONY Xperia Z5 Premium
#> 3:                   Sony 行動微型投影機 MP-CL1
#> 4:                               SONY Xperia Z5
#> 5:                               SONY Xperia Z5
#> 6:            SONY SBH70 (防水IP57)無線藍牙耳機
#>                                                            describe price
#> 1: 送9H玻璃保貼+保護套 Sony Xperia Z5 Compact E5823 4.6吋八核輕旗艦 13880
#> 2:                   ▼送32G卡+手機立架+玻璃貼SONY Xperia Z5 Premium 19900
#> 3:                                       Sony 行動微型投影機 MP-CL1 11900
#> 4: 超值▼送行動電源+保護貼+保護套▼SONY Xperia Z5 5.2吋美型防水旗艦機 16888
#> 5:                           ★送保貼+支架+觸控筆吊飾★SONY Xperia Z5 16990
#> 6:                                SONY SBH70 (防水IP57)無線藍牙耳機  2188
#>    author brand publishDate isPick
#> 1:                               0
#> 2:                               0
#> 3:                               0
#> 4:                               0
#> 5:                               0
#> 6:                               0

非結構化資料解析: Regular Expression

Package: stringr

• str_replace(), str_replace_all()
• str_extract(), str_extract_all()

library(stringr)
fruits <- c("one apple", "two pears", "three bananas")
str_replace(fruits, "[aeiou]", "-")

#> [1] "-ne apple"     "tw- pears"     "thr-e bananas"

str_replace_all(fruits, "[aeiou]", "-")

#> [1] "-n- -ppl-"     "tw- p--rs"     "thr-- b-n-n-s"

shopping_list <- c("apples x4", "bag of flour", "bag of sugar", "milk x2")
str_extract(shopping_list, "[a-z]+")

#> [1] "apples" "bag"    "bag"    "milk"

str_extract_all(shopping_list, "[a-z]+")

#> [[1]]
#> [1] "apples" "x"     
#> 
#> [[2]]
#> [1] "bag"   "of"    "flour"
#> 
#> [[3]]
#> [1] "bag"   "of"    "sugar"
#> 
#> [[4]]
#> [1] "milk" "x"

Exercise : 東森房屋

• 東森房屋
• 試著找到「新北市」的資料

Exercise: 紫外線即時監測資料

行政院環境保護署環境資源資料開放平台提供了一系列的RESTful Api供大家取用，請試著把 紫外線即時監測資料的資料取回來並轉成data.frame

Answer

紫外線即時監測資料

Save your data

• write.csv: write data.frame as csv file
• download.file: save html, jpeg, etc
• writeBin: write binary object into disk
• RSQLite: SQLite connector in R

write.csv

library(jsonlite)
library(httr)
url = "http://ecshweb.pchome.com.tw/search/v3.3/all/results?q=sony&page=1&sort=rnk/dc"
res_df = GET(url) %>% 
    content(res, as = "text") %>% 
    fromJSON() %>% 
    .$prods     # equivelent to (function(x) {x$prods})
write.csv(res_df, "resources/data/pchome.csv", row.names = FALSE)

download.file

To download a file from the Internet. download.file takes advantage of internet utilities such as curl or wget and may fail if you don’t have any of these utilities in your system.

dest_dir = "resources/data/download"
dir.create(dest_dir, showWarnings = FALSE, recursive = TRUE)

# Download whole HTML file
download.file("https://www.r-project.org/", 
              file.path(dest_dir, "r-project.org.html"))

# Download image
download.file("https://www.r-project.org/Rlogo.png",
              file.path(dest_dir, "Rlogo.png"))

list.files(dest_dir)

writeBin

To write binary data to your local disk.

r = GET("http://opendata.epa.gov.tw/webapi/api/rest/datastore/355000000I-000004/?format=json")
# Set as = "raw" to prevent any character encoding
bin = content(r, as = "raw")
writeBin(bin, "resources/data/download/uv.json")

Database

• RSQLite: A light-weight database engine interface in R
  • PChome Example
• Other relational database connector in R
  • RODBC
  • RJDBC
  • RMySQL
  • …

Case study

• 東森房屋
• Guestbook
• Yahoo Stock
• 紫外線即時監測資料
• PTT Gossiping
• Starbucks
• Seven Eleven
• FamilyMart # json & referer
• PChome
• 公開資訊觀測站
• 中央氣象局觀測資料查詢系統

References

• rvest
• httr
• HTTP Methods
• An Introduction to APIs
• Advanced R
• data.table
• RSQLite